intracellular atp can regulate afferent arteriolar tone via atp-sensitive k+ channels in the rabbit

Intracellular ATP Can Regulate Afferent Arteriolar Tonevia ATP-sensitive K+ Channels in the Rabbit

John N. Lorenz, * Jurgen Schnermann, * Frank C. Brosius,6 Josie P. Bnggs, * $ and Philip B. Furspan *Departments of *Physiology and tInternal Medicine, University ofMichigan, Ann Arbor, Michigan 48109

Abstract Introduction

Studies were performed to assess whether ATP-sensitive K+(KATP) channels on rabbit preglomerular vessels can influenceafferent arteriolar (AA) tone. K+ channels with a slope conduc-tance of 258±13 (n = 7) pS and pronounced voltage depen-dence were demonstrated in excised patches from vascularsmooth muscle cells of microdissected preglomerular seg-ments. Channel activity was markedly reduced by 1 mM ATPand in a dose-dependent fashion by glibenclamide (10-9 M to10-6 M), a specific antagonist of KAT channels. 10- M dia-zoxide, a K+ channel opener, activated these channels in thepresence of ATP, and this effect was also blocked by glibencla-mide. To determine the role of these KAT channels in the con-trol of vascular tone, diazoxide was tested on isolated perfusedAA. After preconstriction from a control diameter of 13.1±1.1to 3.5±2.1 gm with phenylephrine (PE), addition of 1O-s Mdiazoxide dilated vessels to 11.2±0.7 um, which was not differ-ent from control. Further addition of 10-5 M glibenclamidereconstricted the vessels to 5.8±1.5 gm (n = 5; P < 0.03). Insupport of its specificity for KATP channels, glibenclamide didnot reverse verapamil induced dilation in a separate series ofexperiments. To determine whether intracellular ATP levelscan effect AA tone, studies were conducted to test the effect ofthe glycolytic inhibitor 2-deoxy-D-glucose. After preconstric-tion from 13.4±3.2 to 7.7±1.3 Mm with PE, bath glucose wasreplaced with 6 mM 2-deoxy-D-glucose. Within 10 min, thearteriole dilated to a mean value of 11.8±1.4 Mum (n = 6; NScompared to control). Subsequent addition of 10-5 M gliben-clamide significantly reconstricted the vessels to a diameter of8.6±0.5 gm (P < 0.04). These data demonstrate that KATpchannels are present on the preglomerular vasculature and thatchanges in intracellular ATP can directly influence afferent ar-teriolar tone via these channels. (J. Clin. Invest. 1992. 90:733-740.) Key words: kidney * glibenclamide * diazoxide * 2-deoxy-i-glucose * patch clamp

This work was presented in part in abstract form at the 45th AnnualFall Conference ofthe Council for High Blood Pressure Research, Chi-cago, IL, 24-27 September, 1991, and at the 24th Annual Meeting ofthe American Society of Nephrology, Baltimore, MD, 17-20 No-vember, 1991.

Address correspondence to Dr. Josie P. Briggs, Department of In-ternal Medicine, Division of Nephrology, 1560 Medical Science Re-search Building II, University of Michigan, Ann Arbor, MI 48109.

Receivedfor publication 7 November 1991 and in revisedform 23March 1992.

ATP-sensitive potassium channels (KATP)' were first describedin cardiac muscle ( 1), and have since been reported and char-acterized in skeletal muscle, pancreatic ,B-cells, vertebrateaxons, and, most recently, in vascular smooth muscle cells (2-5). Since these channels are very sensitive to intracellular con-centrations of ATP, they provide an important potential linkbetween cell metabolism and electrical excitability (2-4). Re-cent evidence suggests that these channels may play an impor-tant role in the pharmacologic and physiologic regulation ofvascular smooth muscle tone. For example, several vasodila-tors known to hyperpolarize vascular smooth muscle cells, not-ably minoxidil, cromakalim, and diazoxide, have been foundto activate ATP-sensitive K+ channels, and the action of thesedrugs can be reversed by the sulfonylurea glibenclamide, whichhas been shown to be a selective inhibitor of KATP channels invascular smooth muscle cells (5-7). KATP channels on vascularsmooth muscle cells therefore may represent an important regu-latory mechanism for the control of vascular tone.

In addition to being sensitive to the intracellular ATP con-centration, KATP channel activity can also be influenced by awide variety of other physiological variables including ADPand other nucleotide concentration, intracellular pH, divalentcation concentration, and G-protein activity (3). There is alsoindirect evidence that phosphorylation ofcertain types ofATP-sensitive K+ channels is necessary to retain the open state (4,8). Finally, evidence strongly suggests that certain endogenousvasodilator peptides, including vasoactive intestinal polypep-tide (5), endothelium derived hyperpolarizing factor (9), andcalcitonin gene related peptide (10), produce relaxation atleast in part by activating these K+ channels. In spite of thissupporting evidence, there is still debate concerning the actualimportance of ATP in controlling the activity of these chan-nels. The inhibitory constant ofATP for these channels is verylow, suggesting that at normal physiologic levels ofintracellularATP channel activity would be completely absent (4). Thus, inspite of the pharmacologic evidence regarding the role of thesechannels, there has been no direct evidence to support the con-cept that actual changes in intracellular ATP concentration caninfluence vascular smooth muscle tone via ATP-sensitive K+channels.

The aim of the present study was to determine whetherKATP channels are present on the preglomerular microvascula-ture ofthe rabbit kidney using both the patch clamp techniqueand the isolated perfused afferent arteriole technique and toassess whether manipulation of cell ATP levels affect vasculartone through these channels. Single channel responses to diaz-

1. Abbreviations used in this paper: KATP, ATP-sensitive potassium;PE, phenylephrine.

A TP-sensitive K+ Channels in the Afferent Arteriole 733

J. Clin. Invest.© The American Society for Clinical Investigation, Inc.0021-9738/92/09/0733/08 $2.00Volume 90, September 1992, 733-740

oxide, a K+ channel opener, and glibenclamide, a specificblocker of KATP channels, were determined on excised mem-brane patches obtained from preglomerular vascular smoothmuscle cells ofthe rabbit kidney. Functional correlates to thesesingle channel events were then assessed by measuring the ef-fects of these agents on the diameter of isolated afferent arteri-oles. To determine whether intracellular ATP can influencepreglomerular resistance by interacting directly with thesechannels, the afferent arteriolar response to intracellular ATPdepletion was tested, as well as the ability of glibenclamide tomodify this response. The studies establish the presence ofATP-sensitive K+ channels on the afferent arteriole and sug-gest that modulation of cell ATP levels may affect renal bloodflow, in part through an effect on these channels.

Methods

Patch clamp studiesAnimals and tissue preparation. All experiments were performed onrenal vascular smooth muscle cells obtained from New Zealand whiterabbits weighing 0.8-1.5 kg and maintained on a standard rabbit chow.After euthanasia, the left kidney was removed, decapsulated, and slicedtransversely through the papilla. The slices were placed in ice-coldDME containing 3% fetal calf serum, and dissection was done at 4°C.The dissection medium was bubbled with 5% CO2 and 95% 02, and itspH was adjusted to 7.4 immediately before use. The outer corticalportions of interlobular vascular trees were closely examined, and ter-minal branches ofan interlobular artery, with several attached afferentarterioles, were dissected free using sharpened forceps. Glomeruli weregently removed using a razor blade. After dissection, vascular segmentswere incubated for 20 min at 32°C in Hanks' solution containing: (inmilligrams per milliliter) collagenase (0.5; 115 U/ml); trypsin inhibi-tor (0.3); papain (0.4; 5.2 U/ml); dithiothreitol (0.3); BSA (7.5); pH,7.4. Specimens were then washed and placed in normal Hanks' solu-tion at room temperature and used within 2 h.

Patch clamp technique. Single channel currents were obtained us-ing standard patch clamp techniques as described by Hamill et al. ( 11 ).All measurements were made at room temperature on excised patchesin the inside-out configuration. Isolated vascular smooth muscle cellswere visualized using an inverted microscope. Micropipettes were fabri-cated from capillary tubes and had a tip resistance of5-10 Mohm. Sealresistances were in the range of 5-20 Gohm. Single channel currentswere recorded through a patch clamp amplifier (8900; Dagan Corp.,Minneapolis, MN). Amplifier output was filtered with an 8-pole Besselfilter set at a corner frequency of 3.4 kHz, digitized, and stored onvideotape.

Bath and pipette filling solutions ofthe following composition wereused for most observations (in millimolars): 145 KCI, 10 Hepes(N-[2-hydroxyethyl] piperazine-N'-[2-ethanesulfonic acid]), 10 glu-cose, and KOH to bring pH to 7.4. A calcium-buffered bath solutionwith the following composition was used for experiments in whichcalcium or ATP levels were varied (free calcium < 10 nM) (millimo-lar): 113 KCI, 2.02 EGTA, 20.2 K2Pipes (Piperazine-N-N'-bis [2-eth-anesulfonic acid]); pH 7.0. Calcium concentration was varied by addi-tion of CaCI2, with free calcium determined from the EGTA dissocia-tion constant. ATP was added as the K salt (Sigma Chemical Co., St.Louis, MO), and free calcium concentration was adjusted by additionof CaCI2, with correction for the ATP-dissociation constant. All drugswere added to the bath solution. Vehicles were tested independently.

Protocol. After formation of the gigaseal to establish the on-cellconfiguration, the pipette tip was passed briefly through the air-liquidinterface to destroy the cell body and leave an inside-out patch ofmem-brane. In this configuration the properties such as conductance, volt-age, and divalent ion dependency, ATP-sensitivity, selectivity of thepotassium channel present were determined. Potassium channels inac-

tivated by ATP were characterized further by their response to bathaddition of the KATP antagonist glibenclamide (Sigma Chemical Co.).in concentrations from 10-' to 10-6 M, and/or the KATp activator,diazoxide (Quad Pharmaceuticals Inc., Indianapolis, IN) at a concen-tration I0-5 M.

Isolated perfused arteriole studiesAnimals and technique. All experiments were performed on isolatedafferent arterioles obtained from New Zealand white rabbits as de-scribed above. After selecting an afferent arteriole and its attached glo-merulus from the outer cortical portion ofan interlobular tree, tubularfragments were carefully removed using sharpened forceps. Care wastaken not to stretch the vessels or to touch the vascular structures at theglomerular hilum. After dissection, the interlobular artery was cut at asite just before the branch point of the afferent arteriole, so that thefinal specimen consisted of a segment of interlobular artery, the affer-ent arteriole, the glomerulus, and a fragment of the efferent arteriole.Inspection ofthe specimen under high power usually revealed the pres-ence ofthe macula densa cells. Specimens were not accepted ifthe totaldissection time exceeded 60 min.

After dissection, specimens were transferred into a thermoregulatedchamber placed on the stage of an inverted microscope (OlympusIMT-2) for cannulation as previously described ( 12). Briefly, usingconcentric pipettes mounted on a moveable track system ( 13), theproximal end of the interlobular artery was aspirated into a holdingpipette. The perfusion pipette was then advanced through the lumen ofthe interlobular artery into the afferent arteriole. The perfusate wasdriven by a hydraulic pressure head of 1 10 cm H20. After establishingvessel perfusion, a small cover slip was placed over the top of the bathcreating a small, closed, and stable chamber with a total volume of- 100 ul. The bath temperature was then elevated to 38°C. Fresh bathmedium was supplied by a syringe pump at a rate of 200 Mil/min andremoved continuously by aspirating the effluent. Initial spontaneouscontractions with varying intensity were observed during the warmupperiod in the majority of vessels. These contractions in general sub-sided during an equilibration period of 60 min. Before studies werestarted, specimens were inspected for cellular integrity and the presenceof macula densa cells using differential interference contrast optics(Olympus IMT-2-DIC) at magnifications of 600.

ProtocolsSeries 1: diazoxide. In five experiments, the ability of diazoxide todilate preconstricted arterioles was tested. In all ofthe following experi-ments, a phenylephrine (PE) dose-response curve for each vessel wasfirst determined with doses from 10-7 to 10-5 M, and a dose ofPE wasselected to preconstrict the vessel by 50-80%. After preconstrictionwith PE, diazoxide was added to the bathing medium at a concentra-tion of 10-5 M. When a stable response to diazoxide had been obtained,usually after 3-4 min, 10' M glibenclamide was also added to thebathing medium to test its ability to counteract the affect ofdiazoxide.After 5 min, all agents were removed from the bath to determine recov-ery. In a separate series of control experiments, the time course of PEconstriction was determined over a 1 0-min period, first in the presenceofdiazoxide vehicle and then in the presence ofglibenclamide vehicle.

Series 2: verapamil. To test the specificity ofthe constrictor effect ofglibenclamide, studies were performed to determine its ability to re-verse the effect of verapamil. After preconstricting arterioles with PE,verapamil was added to the bath at a concentration of 10' M, andsteady-state vasodilation was achieved in - 3 min. Glibenclamide wasthen added to the bath at 10' M for a 5-min period, which was fol-lowed by a recovery period.

Series 3: 2-deoxy-D-glucose. All experiments were performed usingRPMI 1640 glucose-deficient medium rather than DME as the bathingmedium so that glucose could be directly varied. To determine whetherdecreasing intracellular levels ofATP can dilate afferent arterioles, theeffect of replacing glucose with 2-deoxy-D-glucose was tested. Substitu-tion of this analogue for glucose has been shown to decrease the intra-

734 J. N. Lorenz, J. Schnermann, F. C. Brosius, J. P. Briggs, and P. B. Furspan

cellular concentration ofATP (14). Before the beginning ofthis proto-col, the luminal perfusate solution was exchanged for an identical solu-tion without glucose. Afferent arterioles were first preconstricted with ahalf-maximal dose of PE in the presence of 100 mg% glucose. Thebathing medium was then replaced with an identical solution in whichthe glucose had been completely replaced with 100 mg% 2-deoxy-D-glucose. When a stable response had been achieved in the presence ofPE and 2-deoxy-D-glucose, usually within 10 min, 10- M glibencla-mide was then added to the bath for a period of 5 min, which wasfollowed by a final recovery period.

Solutions and reagents. Dissection and bath media were preparedfrom DME containing Ham's nutrient mixture F- 12 with the additionof 1.2 g/liter NaHCO3 (Sigma Chemical Co.). Before use this solutionwas aerated with 95% 02/5% CO2 for 45 min and its pH was adjusted to7.4. When used as dissection medium, fetal calfserum (Gibco Labora-tories, Grand Island, NY) was added to a final concentration of 3%.When used as bath medium, 1 mM CaC12 (final [Ca] = 2 mM) wasadded, and bovine serum albumin (Sigma Chemical Co.) was added toa concentration of 0.5%. No significant difference in bath medium pHwas noted between delivered and collected samples. Vessel perfusionfluid was a modified Krebs-Ringer-HCO3 buffer containing (millimo-lar): 115 NaCl, 25 NaHCO3, 0.96 NaH2PO4, 0.24 Na2HPO4, 5 KCI,1.2 MgSO4, 2 CaC12, and 5.5 glucose. The buffer was gassed with 95%02/5% CO2 and pH adjusted to 7.4 and bovine serum albumin wasadded to a final concentration of 1%. In experiments requiring glucose-free medium (series 3), RPMI 1640-deficient medium without glu-cose (Sigma Chemical Co.) was used instead of DME. Glucose or 2-deoxy-D-glucose was added where necessary, and the medium was oth-erwise treated identically to the DME.

Hanks' solution contained: (millimolar) 140 NaCl, 5.4 KCI, 0.44KH2PO4, 0.42 NaH2PO4, 4.17 NaHCO3, 0.1 CaCl2, 5 Hepes, 5.55glucose; pH adjusted to 7.4 with NaOH. PE (Sigma Chemical Co.),diazoxide (Quad Pharmaceuticals), and 2-deoxy-D-glucose (SigmaChemical Co.), were dissolved in a stock solution of saline and diluted100-fold in the final bath medium. Glibenclamide (Sigma) was dis-solved in DMSO and diluted 1,000-fold in the final bath medium. ATP(Sigma) for patch clamp studies was dissolved in distilled water.

Measurements and statisticsVascular diameter changes were recorded on videotape at a magnifica-tion of 600 using differential interference contrast optics. Intraluminalvascular diameters were measured on the video images using an imageanalysis system (Cue 2; Olympus Corp.). Since the luminal diameter ofthese vessels typically displayed a prominent heterogeneity along theirlength, measurements were made at several sites. Values reported inthe text reflect the diameters measured at the most responsive segmentobserved for each vessel. Measurements were made on image framestaken after a steady-state diameter had been achieved for each treat-ment.

Statistical analysis ofthe data was performed on diameter measure-ments (expressed in micrometers) obtained directly from the imageanalysis system which had been previously calibrated using a slide mi-crometer. In all perfusion studies, statistical analysis was performedusing a single-factor, within-subjects analysis of variance (ANOVA),and the Newman-Keuls multiple range test was used to supplement theanalysis, where necessary, to compare individual means. Values in thetext and figures represent means±SEM and differences were regardedas significant at a probability ofP < 0.05.

Results

Patch clamp studiesAlthough channel activity was rarely seen in the on-cell configu-ration, potassium channel activity of some type appeared in- l/3 of the cells examined upon conversion to the inside-out

configuration. Of the excised patches exhibiting activity, 1/3contained a nonrectifying, large conductance potassium chan-nel with a unitary conductance, measured with 145 mM [K]0/145 mM [K]j, averaging 258±13 pS (n = 7). As shown in Fig.1, the activity ofthese large K+ channels was voltage dependentbetween +20 and -60 mv. In previous reports in vascularsmooth muscle cells somewhat lower conductance estimatesand weaker voltage dependence have been reported; for exam-ple, in mesenteric arteries, KATP channels with a conductancevalue of 135 pS were observed with 60mM [K]0/ 120 mM [K]i(4). The difference in conductance probably reflects condi-tions of measurement, although differences in channel proper-ties in different vascular smooth muscle cell types cannot beexcluded.

As Fig. 2 A illustrates, channel activity in these patches (n= 6) was almost completely inactivated by 1 mM ATP and thiseffect was reversible. Mean current in the presence of 1 mMATP was reduced to 17.8±3.2% ofcontrol (n = 6). The inhibi-tion of channel activity by ATP could be partially reversed by10 ,uM diazoxide as illustrated in Fig. 3 (n = 3). Finally, sepa-rate experiments demonstrated that the activity of these chan-

pA

-20

40 80 120

273 pS

mV

mV

c- JLLd j20

V

-6CL ~ r -1tllL rr.2 c

C-Figure 1. (Top panel) Current-voltage plot for a single K+channel.Unitary conductance for this channel was 273 pS. Excised patchesobtained from enzymatically dissociated vascular smooth muscle cellsfrom the rabbit afferent arteriole were studied in the inside-out con-

figuration. (Bottom panel) Original tracings from an excised mem-brane patch showing voltage dependence at membrane potentials of20, -20, and -60 mv.


ControlAControl

C -;5 pAl

2 SOC1mM ATP

2OmVWash

I!At dI I III" 1-llhI i.1C-- "

B10 giM C82+

~~~~~~~~~~~~~~~~~~.A I

liIr il[r rl1' 1

100 nM Ca2+

'I 9Ii I' 'll'I'I

5 pA _

2 seC

I q I... .- I, it" r -'1ll -,[l 11'If- rlwI "'p-- ---Il l 1-Il-l "ij 1 1

5 PAL2 sac

1 mU ATP 30 mV

C i I . II11111 11

1 mM ATP 4 10 L1M Diazoxido

C-MJFigure 3. Original tracings from an excised membrane patch in theinside-out configuration showing that the K+ channels are openedby diazoxide. Under control conditions, channels are almost con-stantly open (top). Addition of 1 mM ATP to the bath greatly reduceschannel activity (middle tracing: recorded 1 min after addition ofATP). Further addition of 10 ,M diazoxide increases K+ channelactivity (bottom tracing: recorded 1 min after addition of diazoxide).

1 mM ATPVh = 30 mV

I

Figure 2. (A) Original tracings from an excised membrane patch inthe inside-out configuration showing ATP sensitivity of the K+ chan-nel. In the absence of ATP (top), channel activity is high. When 1mM ATP is added to the bath (cytosolic side), channel activity ismarkedly inhibited (middle tracing: recorded 1 min after addition ofATP). After washing, channel activity returns toward control levels(bottom tracing: recorded 2 min after washing). (B) Original tracingsfrom an excised membrane patch in the inside-out configurationshowing the lack of an effect of changing bath Ca++ concentrationon the open probability of the ATP-sensitive K+ channel.

nels could be inhibited by glibenclamide in a concentration-de-pendent manner such that channel activity was nearly abol-ished at 1 gM (Fig. 4; n = 3). This effect was not reversible.

Channel activity was not affected by free calcium concen-tration in the range from < 10 nM to 10 ,uM (see Fig. 2 B). Inaddition, in some excised patches from these cells, a large con-ductance (301±9.3 pS, n = 3)K channel was observed that wascalcium dependent; consistent with previous reports (15),these Ca-sensitive channels were not affected by ATP or 1 ,Mglibenclamide.

Isolated perfused arteriole studiesSeries 1. Diazoxide. The result of a single representative per-fused arteriole experiment with diazoxide is shown in the upperpanel of Fig. 5 (open circles). After constriction with 10-6 Mphenylephrine, addition of l0-' M diazoxide to the superfus-ing bath resulted in a vasodilation to nearly control diameterwithin 3 min. Subsequent application of the sulfonylurea gli-benclamide at 10-5 M reconstricted the arteriole to a diameterof - 30% of control. Diameter returned to control levels afterremoval of all agents. The bottom panel of Fig. 5 shows themean results for five such experiments (solid bars). After pre-constriction of the arterioles from a control level of 13.1 ± 1 .1,um to 3.5±2.1 gm (P < 0.01 ) with phenylephrine, addition of

I0-5 M diazoxide dilated the vessels to a diameter of 11.2±0.7,um (P < 0.01 ), which was not significantly different from con-trol. Further addition of 10-1 M glibenclamide reconstrictedvessels to 5.8±1.5 ,m (P < 0.01 compared to control). Recov-ery diameter was 14.4±1.3 ,um. In separate phenylephrine timecontrol experiments (hatched bars, n = 5), maximum contrac-tion was achieved within the first 2 min of PE application as

Events (104)

Control

C_

1 niU Gllbenclimide

c-6 ,. L

- 1 tM Glibenclamide.,.. ..A LI. L -, -;-X

5 pAL2 sec

10 mV

Current (pA)

Figure 4. (Top panel) Original tracings from an excised membranepatch in the inside-out configuration showing dose-dependent chan-nel sensitivity to glibenclamide. Activity is high during control con-ditions (top). In the presence of 1 nM glibenclamide, K+ channelactivity is markedly reduced (middle tracing: recorded 1 min afteraddition of 1 nM glibenclamide). In the presence of I gM glibencla-mide K+ channel activity is nearly absent (bottom tracing: recorded1 min after addition of 1 MM glibenclamide). Adjacent to each re-cording is the current amplitude histograms derived from each 10-speriod shown.


r III A 11111,111

A.

c-

Phenylephrlne 10-6 M

Dlazoxide 10-5 M

Glib 10-5 M

E-

0

0E._

00

:1.

0E0

0

0n0

Time (min)

20

la

-I 1f-->6

16 NOPE time-controlPO5)14

12

0 10

8

6

4

0 2

0CONTROL PE RECOVERY

Diazoxide

Glib

Figure 5. Effect of diazoxide and glibenclamide on afferent arteriolespreconstricted with phenylephrine. (Top) Closed circles represent theresults a single experiment in which an afferent arteriole was seriallyexposed to PE, diazoxide, and glibenclamide in the bath. Open circlesrepresent a single experiment in which an afferent arteriole was ex-posed to PE alone for the duration of the experiment; diazoxide andglibenclamide vehicle were added at appropriate times. (Bottom)Mean stable diameters for five diazoxide/glibenclamide experiments(solid bars) and for five PE time-control experiments (hatched bars).*JP < 0.01 compared to control period in the same group.

diameter decreased from 16.0±1.9 Am to 2.0± 1.0 Am (P< 0.01 ). During application of diazoxide vehicle, the PE con-traction waned slightly, but not significantly, as diameter in-creased to 5.8±1.9 gim. Thereafter, the diameter remainedstable at 5.5±1.4 ,um during glibenclamide vehicle application,and diameter returned to 15.3±1.7 ,um during recovery. Itshould be noted that the vessel diameter during the experimen-tal glibenclamide period was not significantly different fromthat during the glibenclamide vehicle period (5.8±1.5 vs5.5±1.4 ,um). A representative experiment showing the mainte-nance of the phenylephrine response is shown by the opencircles in the upper panel of Fig. 5.

Series 2. Verapamil. The upper panel in Fig. 6 illustratesthe results from a single verapamil experiment. Following con-striction with phenylephrine, addition of l0' M verapamilresulted in vasodilation to control values, and this effect wasnot further modified by the addition of glibenclamide. Themeans of six of these experiments are illustrated in the lowerpanel of Fig. 6. Arterioles were constricted from 15.2±1.0 Mmto 2.7± 1.1 Mum (P < 0.001 ) with PE, and subsequent addition

Phenylephrln. 5 x 10.6 U

Verepamil 10-5 U

Glib 10-5 M

Time (min)

16

CONRO _4ERCVR

12

0 I10E0 8

6

0 4

0n 2

0CONTROL PE RECOVERY

Verapa ml

Glib

Figure 6. Effect of verapamil and glibenclamide on afferent arteriolespreconstricted with phenylephrine. (Top) Results ofa single experi-ment in which an afferent arteriole was serially exposed to PE, vera-pamil, and glibenclamide in the bath. (Bottom) Mean stable diame-ters for six verapamil/glibenclamide experiments. *P < 0.001 com-pared to control period.

of 10- M verapamil dilated vessels to 14.8±1.2 ,um, which isnot significantly different from control. Subsequent additionof l0' M glibenclamide did not effect vessel diameter,which then averaged 15.1±1.0 ,um. Recovery diameter was15.9±1.3 ,im.

Series 3. 2-deoxy-D-glucose. The upper panel of Fig. 7shows the results from a single experiment with 2-deoxy-D-glu-cose. After preconstriction with phenylephrine, exposure ofthevessel to 100 mg% 2-deoxy-D-glucose in the absence ofglucoseresulted in a gradual vasodilation over a period of 10 min. Thisdilation was partially reversed by the subsequent addition of10-' M glibenclamide. Photomicrographs taken from one ofthese experiments are shown in Fig. 7 B, and letters (a-e) oneach panel correspond to the letters labeling each treatmentperiod in the upper panel of Fig. 7 A. The results of six suchexperiments are shown in the lower panel of Fig. 7. Phenyleph-rine constricted vessels from a mean control diameter of13.4±1.3 Am to 7.7±1.3 gAm (P < 0.01). Substitution of glu-cose with 2-deoxy-D-glucose caused a gradual dilation whichwas complete within 10-15 min. After 3-5 min of exposure to2-deoxy-D-glucose, the mean diameter had decreased onlyslightly to 6.7±2.0 ,im (NS). After 6-9 min, diameter had in-creased to 9.3±2.0 ,im, but this change was also not significant.After 10-15 min, the diameter had increased to 11.7±1.4 ,imwhich was significantly greater than the PE-preconstriction


ba Phe.y!ephrlne 3 x 10-6 M

0 glucoses1OOmg%. 2-deoxy-D-glucos*

Gl!b 10-5 M

4* w ~~~~~~~~~~~~~~~~~~~~~~~~i

~I

Time rnir.

* **

I0 glulcose 5r 2 Iet.xye D glucos

G b

Figure 7. (a) Effect of glycolytic inhibition with 2-deoxy-D-glucose on preconstricted arterioles. (Top) Results of a single experiment in which an

arteriole was serially exposed to PE, 0 glucose/6 mM 2-deoxy-D-glucose, and glibenclamide in the bath. Letters a-e refer to photomicrographsshown in Fig. 8. (Bottom) Mean diameters for six 2-deoxy-D-glucose/glibenclamide experiments. The PE + 2-deoxy-D-glucose period lasted10-15 min and mean diameters were calculated at the ½/3 (3_-5'), 2/3 (6'-9'), and final ( l0'-15') time point within that period. *P < 0.01, and*P < 0.05 compared to control. (b) Photomicrograph showing the effects of glycolytic inhibition and of glibenclamide on preconstricted afferentarterioles. Letters a-e correspond to the labeled time points in the top panel of Fig. 7: a, control; b, 5 X10-6 M PE; c, PE + 0 glucose and 6mM 2-deoxy-D-glucose; d, PE + 0 glucose and 6 mM 2-deoxy-D-glucose + 10-5 M glibenclamide; e, recovery.

level (P < 0.05) and not different from control. Further addi-tion of 10-5 M glibenclamide caused a vasoconstriction to a

mean diameter of 8.0±0.6 ,um (P < 0.05 compared to the pre-vious period). Recovery diameter was 13.9±1.3 ,m.

Discussion

ATP-sensitive K+ channels, first described in cardiac myocytes( 1 ), have now been demonstrated in a wide variety ofcell types(2-4). Although the properties of ATP-sensitive K+ channelsappear to vary somewhat between cell types, all ATP-sensitiveK+ channels are highly selective for potassium and are inhib-ited by intracellular ATP and by sulfonylureas such as gliben-clamide (3). Recent studies have established that channels inthis family are also present on vascular smooth muscle cells.Using patch clamp methods on rabbit mesenteric arterysmooth muscle cells, Standen and co-workers (5) documentedlarge conductance K+ channels ( 135 pS in 60mM [K++]0/ 120mM [K+]j) which were inactivated either by 1 mM ATP or 20,uM glibenclamide and activated by 1 gM cromakalim. ATP-sensitive K+ channels have also been demonstrated by patchclamp studies on smooth muscle cells from rat tail arteries( 16). Sulfonylureas have been shown to reverse the vasodila-

tion produced by agents such as diazoxide, minoxidil, croma-kalim, and pinacidil, supporting the conclusion that theseagents act in part by opening KATP channels (4, 6, 7).

The results of the present study extend these observationsto the resistance vessels of the rabbit kidney. In patch clampstudies using enzymatically dissociated cells from microdis-sected preglomerular vessels of the rabbit kidney, we detectedthe presence of a K+ channel with a unitary conductance of

260 pS under symmetrical K+ concentrations of 150 mM.

These K+ channels were inactivated by mM ATP and byglibenclamide in a dose-dependent fashion from 10-9 to 10-6M. In addition, single channel recordings showed that, in thepresence of millimolar ATP, application of l0-5 M diazoxidecan reopen these K+ channels, providing direct evidence forthe molecular mechanism of this agent. These data, therefore,provide electrophysiological evidence for the presence of a po-tassium channel on the preglomerular vasculature ofthe rabbitwith physiologic and pharmacologic properties consistent witha ATP-sensitive K+ channel.

Functional support for a role of these channels in the con-

trol of preglomerular resistance was obtained in experimentsusing isolated perfused afferent arterioles preconstricted withphenylephrine. The afferent arteriole was used for study be-


R._ ..

a)

Cl>

E

E0

0)

0)

4

2

(

8

6

4

I

"t

cause it is a resistance vessel critical for regulation of renalblood flow and glomerular filtration rate. In the first experimen-tal series, addition of the K+ channel opener, diazoxide, pro-duced vasodilatation which could be reversed by addition ofglibenclamide. In the second experimental series, glucose up-take and glycolytic production ofATP were inhibited by bath-ing perfused vessels in glucose-free medium containing 2-deoxy-D-glucose; this produced a vasodilatation which couldalso be reversed by addition of glibenclamide. In separate con-trol experiments, phenylephrine produced sustained vasocon-striction, unaffected by the application of the drug vehicles,and vasodilation produced by verapamil was not modified byglibenclamide, supporting the specificity of this agent. Takentogether with the patch clamp studies, these data suggest thatactivation of KATP channels results in cell hyperpolarizationand vasodilation, and permit the conclusion that ATP-sensi-tive K+ channels can contribute significantly to the control ofvascular smooth muscle tone in the rabbit afferent arteriole.Previous studies have provided evidence that KAT channelsplay a role in hypoxic vasodilation in coronary vessels (17).Our observations suggest that variation in glucose availabilityin intact vascular smooth muscle cells also influences vasculartone through an effect on ATP-sensitive K+ channels.

Patch clamp studies have reported that the inhibitory con-stant for KATP channels in inside-out patches from vascularsmooth muscle is between 10 and 200 uM (3), and the patchclamp results reported here are consistent with this estimate.Since cytosolic ATP concentrations are normally quite high, inthe millimolar range, this low inhibitory constant would pre-dict channel closure under resting conditions. In the presentstudy two observations suggest that ATP-sensitive K+ channelson the vascular smooth muscle cells were largely closed or inac-tive during resting conditions. First, channel activity could notbe detected before cell detachment in the patch clamp experi-ments, and second, treatment with glibenclamide in restingafferent arterioles had no vasoactive effect (unreported obser-vation). The acute vasodilation observed in the presence of2-deoxy-D>glucose appears at first somewhat surprising be-cause such manipulation is unlikely to lower average intracel-lular ATP concentration below the reported inhibitory con-stant ( 14, 18). There are several possible explanations for thisapparent discrepancy. ATP may not be the sole factor in deter-mining KATP channel activity. In both ,8-cells ( 19) and cardiacmuscle cells (20), the inhibitory constant of ATP on thesechannels is significantly increased in the presence ofADP, sug-gesting that the ATP:ADP ratio may actually determine chan-nel activity in the intact cell. In support ofthis notion, a recentstudy reported that, in cerebral microvessels, glycolysis is re-quired to maintain normal values ofboth ATP and ATP:ADPratio ( 18 ). Channel activity has also been reported to be sensi-tive to cell pH, and the levels of other nucleotides (21-23).Furthermore, as a number of investigators have observed, thehigh conductance of these channels and their relative abun-dance on many cell types has the consequence that smallchanges in the relative open probability may have substantialimpact on cell membrane potential (4, 24, 25).

An additional factor, which may be particularly relevant forvascular smooth muscle cells, is that levels of ATP near theplasma membrane may not be the same as those in other intra-cellular locations. Such heterogeneity may arise as a result ofthe compartmentation of glucose metabolism described in the

vascular smooth muscle cell (26). Vascular smooth musclecells exhibit a distinctive set of characteristics, largely eluci-dated by work on muscle energetics, which make it likely thatthe availability ofglucose has an important effect on cell func-tion. First, intracellular glucose concentration in vascularsmooth muscle is very low (27), so that glucose transportwould appear to be rate limiting in intracellular glucose metab-olism. Second, the energy metabolism ofvascularsmooth mus-cle is unusual in that the rate of glycolysis is substantial evenunder aerobic conditions. Approximately 25% ofATP synthe-sis is provided by the glycolytic breakdown of glucose (28).Third, ATP and glucose metabolism in vascular smooth mus-cle shows extensive compartmentation. The contractile func-tion of the cell appears to be primarily fueled by oxidativemetabolism, while membrane functions are dependent on gly-colytic metabolism (26, 28). Thus, glycolytic inhibitors such as2-deoxy-D-glucose, which, in the absence of glucose, lowersATP levels in a variety of cells (29) including vascular smoothmuscle cells ( 14, 30, 31 ), would be expected to have a predomi-nant effect on membrane related functions. In support of thisnotion, Weiss and co-workers have reported that ATP-sensi-tive K+ channels in cardiac myocytes are more closely depen-dent on glycolytic than on oxidative metabolism (32). Itshould be noted that treatment with 2-deoxy-D-glucose maylead to a number ofrelated metabolic effects that can influencetone independent of the effect on ATP-sensitive K+ channels.For example, low ATP levels might alter intracellular Ca++through a direct effect on Ca++ channels (33 ), it might directlyaffect the rate of myosin light chain kinase phosphorylation( 31 ), and it might directly effect the release of vasoactive sub-stances from the endothelium (34). The demonstration thatthe vasodilation produced by 2-deoxy-D-glucose could bepromptly reversed by addition ofthe specific ATP-sensitive K+channel inhibitor glibenclamide, supports the participation ofKAT channels in the change in vessel tone. These observationsare consistent with the hypothesis that reduction in the levels ofglycolysis-derived ATP in the vascular smooth muscle cell caninfluence tone through a direct effect on KATP channels. Sinceour perfused vessel preparations include both endothelial cellsand vascular smooth muscle cells, an alternate possibility isthat removal of glucose and substitution of 2-deoxy-D-glucoseresulted in release from the endothelial cell of a factor whichactivated the vascular KATP channels. There is some evidencethat endothelial cells produce a hyperpolarizing factor actingthrough this class of channels (5, 35). In the only studies toaddress directly the effects of2-deoxy-D-glucose on endothelialfactors, however, this agent applied to rabbit aorta reducedrelease of endothelial derived relaxing factor, and produceddirectionally opposite changes in vascular tone from that pre-dicted by this proposal (34).

In summary, the present experiments demonstrate the pres-ence of ATP-sensitive potassium channels on vascular smoothmuscle cells from small caliber resistance vessels of the rabbitkidney. The properties of these channels are similar to thosepreviously reported from patch clamp studies on larger arteries(5, 10). Functional experiments using the isolated perfusedrabbit afferent arteriole showed that these channels can play arole in the control of arteriolar tone, and demonstrate for thefirst time that manipulation ofglucose availability alters vascu-lar tone through a direct effect on ATP-sensitive K+ channels.These data support the hypothesis that ATP-sensitive K+ chan-


nels participate in the metabolic control ofpreglomerular resis-tance and, by inference, of glomerular filtration rate and renalblood flow.

Acknowledgments

This work was supported by National Institute of Diabetes and Diges-tive and Kidney Diseases grants DK37448, DK-39255, and DK-40042, and National Heart, Lung and Blood grant HL-18575. J. N.Lorenz is the recipient of an Individual Research Service Award DK-08411. F. C. Brosius was supported by DK-01965.

References

1. Noma, A. 1983. ATP-regulated K+ channels in cardiac muscle. Nature(Lond.). 305:147-148.

2. Ashcroft, F. M. 1988. Adenosine 5'-triphosphate-sensitive potassium chan-nels. Annu. Rev. Neurosci. 11:97-118.

3. Ashcroft, S. J. H., and F. M. Ashcroft. 1990. Properties and functions ofATP-sensitive K-channels. Cell. Signalling. 2:197-214.

4. Nelson, M. T., J. B. Patlak, J. F. Worley, and N. B. Standen. 1990. Calciumchannels, potassium channels, and voltage dependence of arterial smooth muscletone. Am. J. Physiol. 259 (Cell Physiol. 28):C3-C18.

5. Standen, N. B., J. M. Quayle, N. W. Davies, J. E. Brayden, Y. Huang, andM. T. Nelson. 1989. Hyperpolarizing vasodilators activate ATP-sensitive K+channels in arterial smooth muscle. Science (Wash. DC). 245:177-180.

6. Winquist, R. J., L. A. Heaney, A. A. Wallace, E. P. Baskin, R. B. Stein,M. L. Garcia, and G. J. Kaczorowski. 1989. Glyburide blocks the relaxationresponse to BRL 34915 (cromakalim), minoxidil sulfate, and diazoxide in vascu-lar smooth muscle. J. Pharmacol. Exp. Ther. 248:149-156.

7. Quast, U., and N. S. Cook. 1989. In vitro and in vivo comparison oftwo K+channel openers, diazoxide and cromakalim, and their inhibition by glibencla-mide. J. Pharmacol. Exp. Ther. 250:261-27 1.

8. de Weille, J. R., M. Fosset, C. Mourre, H. Schmid-Antomarchi, H. Ber-nardi, and M. Lazdunski. 1989. Pharmacology and regulation of ATP-sensitiveK+ channels. Pfluegers Arch. Eur. J. Physiol. 414(Suppl. 1):S80-S87.

9. Brayden, J. E. 1990. Membrane hyperpolarization is a mechanism ofendo-thelium-dependent cerebral vasodilation. Am. J. Physiol. 259 (Heart Circ. Phys-iol. 28):H668-H673.

10. Nelson, M. T., Y. Huang, J. E. Brayden, J. Heschler, and N. B. Standen.1990. Arterial dilations in response to calcitonin gene-related peptide involveactivation of K+ channels. Nature (Lond.). 344:770-773.

1 1. Hamill, 0. P., A. Marty, E. Neher, B. Sakman, and F. J. Sigworth. 1981.Improved patch-clamp techniques for high resolution current recording fromcells and cell-free membrane patches. Pfluegers Arch. Eur. J. Physiol. 391:85-100.

12. Weihprecht, H., J. N. Lorenz, J. P. Briggs, and J. Schnermann. 1991.Vasoconstrictor effect of angiotensin and vasopressin in isolated rabbit afferentarterioles. Am. J. Physiol. 261 (Renal Fluid Electrolyte Physiol. 30):F273-F282.

13. Greger, R., and W. Hampel. 1981. A modified system for in vitro perfu-sion of isolated renal tubules. Pfluegers Arch. Eur. J. Physiol. 389:175-176.

14. Barron, J. T., S. J. Kopp, J. P. Tow, and J. V. Messer. 1989. Effect ofaltering carbohydrate metabolism on energy status and contractile function ofvascular smooth muscle. Biochim. Biophys. Acta. 976:42-52.

15. Langton, P. D., M. T. Nelson, Y. Huang, and N. B. Standen. 1991. Block

of calcium-activated potassium channels in mammalian arterial myocytes bytetraethylammonium ion. Am. J. Physiol. 260:H927-934.

16. Furspan, P. B., and R. C. Webb. 1990. Potassium channels and vascularreactivity in genetically hypertensive rats. Hypertension (Dallas). 15:687-691.

17. Daut, J., W. Maier-Rudolph, N. von Beckerath, G. Mehrke, K. Guenther,and L. Goedell-Meinen. 1990. Hypoxic dilation ofcoronary arteries is mediatedby ATP-sensitive potassium channels. Science (Wash. DC). 247:1341-1344.

18. Gaposchkin, C. G., K. Tornheim, I. Sussman, N. B. Ruderman, and A. L.McCall. 1990. Glucose is required to maintain ATP/ADP ratio ofisolated bovinecerebral microvessels. Am. J. Physiol. 258 (Endocrinol. Metab. 21):E543-E547.

19. Kakei, M., R. P. Kelly, S. J. H. Ashcroft, and F. M. Ashcroft. 1986. TheATP-sensitivity of K' channels in rat pancreatic #-cells is modulated by ADP.FEBS (Fed. Eur. Biochem. Soc.) Lett. 208:63-66.

20. Findlay, I. 1988. Effects ofADP upon the ATP-sensitive K' channel in ratventricular myocytes. J. Membr. Biol. 101:83-92.

21. Lederer, W. J., and C. G. Nichols. 1989. Nucleotide modulation of theactivity of rat heart ATP-sensitive K' channels in isolated membrane patches. J.Physiol. 419:193-21 1.

22. Davies, N. W. 1990. Modulation of ATP-sensitive K' channels in skeletalmuscle by intracellular protons. Nature (Lond.). 343:375-377.

23. Kirsch, G. E., J. Codina, L. Bimbaumer, and A. M. Brown. 1990. Cou-pling of ATP-sensitive K' channels to Al receptors by G proteins in rat ventricu-lar myocytes. Am. J. Physiol. 259:H820-H826.

24. Cook, D. L., L. S. Satin, M. L. J. Ashford, and C. N. Hales. 1988. ATP-sensitive K' channels in pancreatic beta-cells: spare channel hypothesis. Dia-betes. 37:495-498.

25. Nichols, C. G., C. Ripoll, and W. J. Lederer. 1991. ATP-sensitive potas-sium channel modulation ofthe guinea pig ventricular action potential and con-traction. Circ. Res. 68:280-287.

26. Lynch, R. M., and R. J. Paul. 1987. Compartmentation ofglycolytic andglycogenolytic metabolism in vascular smooth muscle. Science (Wash. DC).222:1344-1346.

27. Lynch, R. M., and R. J. Paul. 1984. Glucose uptake in porcine carotidartery: relation to alterations in active Na+-K+ transport. Am. J. Physiol. 247(Cell Physiol. 16):C433-C440.

28. Paul, R. J. 1989. Smooth muscle energetics. Annu. Rev. Physiol. 51:331-349.

29. Webb, J. L. 1966. Effects of 2-deoxy-D-glucose on carbohydrate metabo-lism. In Enzymes and Metabolic Inhibitors. Vol. 11. J. L. Webb, editor. AcademicPress, New York. pp. 386-403.

30. Cremer-Lacuara, M. G., J. L. Lacuara, M. F. de Cuneo, and R. D. Ruiz.1980. Substrate supply and function ofisolated smooth muscle under anoxia andmetabolic inhibition. Can. J. Physiol. Pharmacol. 58:723-730.

31. Barron, J. T., K. Barany, and S. J. Kopp. 1989a. Effects ofATP reductionon the pattern of force development and myosin light chain phosphorylation inintact arterial smooth muscle. Biochim. Biophys. Acta. 1010:278-282.

32. Weiss, J. N., and S. T. Lamp. 1987. Glycolysis preferentially inhibitsATP-sensitive K' channels in isolated guinea pig myocytes. Science (Wash. DC).238:67-69.

33. Ohya, Y., and N. Sperelakis. 1989. ATP regulation of the slow calciumchannels in vascular smooth muscle cells of guinea pig mesenteric artery. Circ.Res. 64:145-154.

34. Weir, C. J., I. F. Gibson, and W. Martin. 1991. Effects of metabolicinhibitors on endothelium-dependent and endothelium-independent vasodila-tion of rat and rabbit aorta. Br. J. Pharmacol. 102:162-166.

35. Taylor, S. G., and A. H. Weston. 1988. Endothelium-derived hyperpolar-ized factor: a new endogenous inhibitor from the vascular endothelium. TrendsPharmacol. Sci. 9:272-274.


intracellular atp can regulate afferent arteriolar tone via atp-sensitive k+ channels in the rabbit

Documents